A community-based randomized controlled trial of iron and zinc
supplementation in Indonesian infants: interactions between iron
and zinc1–3
Torbjörn Lind, Bo Lönnerdal, Hans Stenlund, Djauhar Ismail, Rosadi Seswandhana, Eva-Charlotte Ekström, and Lars-Åke Persson
ABSTRACT
Background: Combined supplementation with iron and zinc dur-
ing infancy may be effective in preventing deficiencies of these
micronutrients, but knowledge of their potential interactions when
given together is insufficient.
Objective: The goal was to compare the effect in infants of com-
bined supplementation with iron and zinc and of supplementation
with single micronutrients on iron and zinc status.
Design: Indonesian infants (n = 680) were randomly assigned to
daily supplementation with 10 mg Fe (Fe group), 10 mg Zn (Zn
group), 10 mg Fe + 10 mg Zn (Fe+Zn group), or placebo from 6
to 12 mo of age. Venous blood samples were collected at the start
and end of the study. Five hundred forty-nine infants completed
the supplementation and had both baseline and follow-up blood
samples available for analysis.
Results: Baseline prevalences of anemia, iron deficiency anemia
(anemia and low serum ferritin), and low serum zinc (< 10.7 mol/L)
were 41%, 8%, and 78%, respectively. After supplementation, the
Fe group had higher hemoglobin (119.4 compared with 115.3 g/L;
P < 0.05) and serum ferritin (46.5 compared with 32.3 g/L;
P < 0.05) values than did the Fe+Zn group, indicating an effect of
zinc on iron absorption. The Zn group had higher serum zinc
(11.58 compared with 9.06 mol/L; P < 0.05) than did the placebo
group. There was a dose effect on serum ferritin in the Fe and
Fe+Zn groups, but at different levels. There was a significant dose
effect on serum zinc in the Zn group, whereas no dose effect was
found in the Fe+Zn group beyond 7 mg Zn/d.
Conclusion: Supplementation with iron and zinc was less effica-
cious than were single supplements in improving iron and zinc
status, with evidence of an interaction between iron and zinc when
the combined supplement was given. Am J Clin Nutr
2003;77:883–90.
KEY WORDS Iron, zinc, infants, randomized controlled trial,
anemia, Indonesia, micronutrient supplementation
INTRODUCTION
Infant and child undernutrition is usually associated with
micronutrient deficiencies that significantly contribute to the
global burden of disease in childhood. Iron deficiency and iron
deficiency anemia have long been a concern (1), and zinc defi-
ciency has become a priority more recently (2). Estimates of the
global burden of disease for iron deficiency anemia in children are
Am J Clin Nutr 2003;77:883–90. Printed in USA. © 2003 American Society for Clinical Nutrition
Downloaded from https://academic.oup.com/ajcn/article-abstract/77/4/883/4689762 by guest on 12 March 2019
based on deaths from severe anemia and on disability related to
cognitive impairment (3). A crude estimate of zinc deficiency in
children indicates that it ranks before iron deficiency anemia but
after vitamin A deficiency in the global burden of diseases and
disabilities (4). To date, no programs have been implemented to
address iron and zinc deficiencies in children. Such programs can
either be food-based, including fortification of commonly used
foods, or consist of the provision of supplements. In low-income
countries, complementary foods for older infants and young chil-
dren are rarely manufactured at a central location, making effi-
cient food fortification difficult to achieve. Supplementation may
therefore be a more realistic alternative.
Iron deficiency anemia is frequently the result of low intakes
of dietary iron, low intakes of meat, and high intakes of phytate,
an inhibitor of iron absorption. An iron-deficient infant popula-
tion commonly also has a low dietary intake of zinc, and the lack
of meat and the high phytate intake result in a low bioavailability
of zinc (5). It has therefore become evident during the past decade
that deficiencies of iron and zinc coexist, and that vulnerable
groups may benefit from iron as well as zinc supplementation.
Zinc intervention trials in infants and children have shown signi-
ficant improvement in growth (6) and decreased morbidity, par-
ticularly from diarrheal disease (7) and malaria (8), although more
recent studies have challenged the latter finding (9, 10).
It is consequently reasonable to believe that supplements con-
taining both iron and zinc would be of value in populations with
1
From the Department of Public Health and Clinical Medicine, Epidemiol-
ogy, Umeå University, Umeå, Sweden (TL, HS, E-CE, and L-AP); the Depart-
ment of Nutrition, University of California, Davis (BL); Community Health
and Nutrition Research Laboratories, Faculty of Medicine, Gadjah Mada Uni-
versity, Yogyakarta, Indonesia (DI and RS); and the International Centre For
Diarrhoeal Disease Research, Bangladesh, Centre for Health and Population
Research, Dhaka, Bangladesh (E-CE and L-AP).
2
Supported by grants from the Swedish Agency for Research Co-operation
with Developing Countries, the Swedish Medical Research Council, the
Swedish Foundation for International Co-operation in Research and Education,
the Swedish Medical Society, the Maud and Birger Gustavsson Foundation, and
the Umeå University Foundation.
3
Address reprint requests to T Lind, Department of Public Health and Clin-
ical Medicine, Epidemiology, Umeå University, SE-901 85 Umeå, Sweden.
E-mail: torbjorn.lind@epiph.umu.se.
Received March 1, 2002.
Accepted for publication August 28, 2002.
883
884 LIND ET AL
high-phytate diets and low meat intakes. However, it has been shown
that a high intake of nonheme iron can inhibit the absorption of zinc
(11–13), most likely by competition for absorptive pathways. Simi-
larly, it has been shown that a high ratio of zinc to iron can inhibit
iron absorption (14, 15). These interactions were found when the
micronutrients were given in a water solution, but not when given in
meals or in infant formulas (13, 15–18). Thus, it is possible that iron
and zinc, when given together as a supplement, may interact with
each other and compete for absorption in the small intestine.
We performed a community-based, randomized, double blind,
placebo-controlled trial with a factorial design. Our hypothesis
was that daily iron treatment and iron combined with zinc treat-
ment for infants from 6 to 12 mo of age would result in both
higher hemoglobin concentrations and higher serum ferritin
(SF) values than would treatment with placebo. Furthermore,
we postulated that this effect would be stronger in the iron
group than in the combined iron and zinc supplementation
group. Similarly, we postulated that both zinc treatment and iron
combined with zinc treatment would result in higher serum zinc
concentrations than would treatment with placebo, and that this
effect would be stronger in the zinc only group than in the com-
bined treatment group. Finally, we explored sex differences in
the main outcomes.
SUBJECTS AND METHODS
Setting
In Central Java, Indonesia, childhood malnutrition is a public health
problem. Stunting reportedly affects some 40% of children under 5 y
of age (19), and anemia is prevalent, among both women and children
(20). Breast-feeding is common and of long duration, but exclusive
breast-feeding is rare and complementary foods are introduced early.
The plant-based diet of the area contains little animal protein, and low
amounts of iron and zinc together with large amounts of phytate ren-
der the diet inadequate in terms of iron and zinc and places the infants
at high risk of developing micronutrient deficiencies (21).
Participants
The study was conducted from July 1997 to May 1999 in Pur-
worejo, a predominantly agricultural district in the southern part
of Central Java. The district is the location of a health surveillance
project run by the Community Health and Nutrition Research Lab-
oratories (CHN-RL) of Gadjah Mada University. Infants were
recruited from the surveillance system, with a maximum of 50
infants being assessed for eligibility per month. For feasibility rea-
sons, only infants who lived near the health centers were recruited.
Healthy, singleton infants from the surveillance area who were
aged ≤ 6 mo and whose mothers had been monitored during preg-
nancy and birth were considered eligible for enrollment. Children
with metabolic or neurologic disorders; physical handicaps affect-
ing development, feeding, or activity; or severe or protracted ill-
ness and infants with hemoglobin concentrations < 90 g/L on
assessment of eligibility were excluded after examination by the
study physician (RS). Infants in whom initial blood sampling failed
and in whom hemoglobin could not be assessed were excluded.
Written informed consent was required for inclusion. The study was
approved by the Ethical Committee of Biomedical Research involving
Human Subjects, Faculty of Medicine, Gadjah Mada University,
Yogyakarta, Indonesia, and the Research Ethics Committee, Faculty of
Medicine and Odontology, Umeå University, Umeå, Sweden.
Interventions
Infants were randomly assigned to 1 of 4 treatment groups: iron
(Fe group), zinc (Zn group), iron + zinc (Fe+Zn group), or
placebo. Supplements were administered by the parents or care-
takers once daily. Fieldworkers oversaw and administered the
daily dose every third day and monitored the intake the previous
2 d by means of parent recall. These fieldworkers also monitored
infant morbidity (not reported here). Empty bottles were replaced
every 2 wk and the remaining syrup, if any, was measured and reg-
istered. The supplementation was initiated at 6 mo of age and con-
tinued until the child reached 12 mo of age (180 d of supplemen-
tation). Food intake was monitored with monthly 24-h recalls but
was not subjected to intervention.
Downloaded from https://academic.oup.com/ajcn/article-abstract/77/4/883/4689762 by guest on 12 March 2019
The 4 supplements provided the infants with a daily dose of either
10 mg Fe as ferrous sulfate, 10 mg Zn as zinc sulfate, 10 mg of both
Fe and Zn, or placebo in a sweet-tasting syrup. All supplements
included, per dose (1.6 mL, ie, 2 measuring pipettes), 30 mg ascor-
bic acid, sugar, and water. PT Konimex, Solo, Indonesia, manufac-
tured the supplements in association with the Department of Phar-
macology, Gadjah Mada University. The supplements were tested
for taste acceptance in the study area before initiation of the study.
Outcomes
Primary outcomes for the trial were hemoglobin concentrations
and indicators of iron [SF and serum transferrin receptor (sTfR)]
and zinc (serum zinc) status at the 12-mo endpoint of the inter-
vention. In addition, serum copper concentrations were analyzed.
Other functional outcomes (not reported here) were physical
growth, morbidity, and psychomotor development.
Sample size
Sample size calculations were based on the primary outcomes of
physical growth (knee-heel difference, 1 mm in 3 mo), psychomo-
tor development (Bayley scales of development, 5-point differ-
ence), and diarrheal disease morbidity (relative risk: 0.65). One
hundred seventy infants per group would also allow for a dropout
rate of 20% and provide 136 infants for analysis. This group size
was estimated to allow for detection in differences of hemoglobin
of 2.5 g/L ( = 0.05, power = 80%), whereas 5 g/L was considered
a clinically significant difference. For SF and serum zinc, no judg-
ments about clinically significant differences were made.
Randomization
Randomization was planned and generated by an independent
statistician, and was performed in blocks of 20. The pharmaceu-
tical company marked the 4 different supplements with letter
codes, blinded to researchers and participants. Information on
group assignment was kept in a safe at the administrative offices
of Gadjah Mada and Umeå Universities until after the intent-to-treat
analysis. Participants were assigned to treatment groups by the
recruitment field staff, who strictly followed the randomization
list. The laboratory assessing the biochemical outcomes was not
aware of the randomization groups.
Baseline and follow-up data collection
Trained fieldworkers collected socioeconomic information in
home interviews. Fieldworkers or community midwives meas-
ured birth weight at the time of delivery. Anthropometric meas-
urements (nude weight, recumbent length, and knee-heel length)
were taken monthly from 4 to 12 mo of age (6- and 12-mo assess-
ments are reported here), as were 24-h diet recalls. After the end of
 IRON AND ZINC SUPPLEMENTATION IN INFANTS
FIGURE 1. Trial profile.
the supplementation period, the families were interviewed on per-
ceived side effects (abdominal pain, decreased appetite, vomiting, diar-
rhea, constipation, and increased crying or fussiness) as well as the
general health condition of the child before and after supplementation.
At the time of collection of the endpoint blood samples, body tem-
perature was measured and a 2-wk recall of morbidity was obtained.
Biochemical assays
Venous blood was collected from the study subjects before and
after the supplementation period, ie, at 6 and 12 mo of age, with the
use of a zinc-free vacuum system. A drop of blood was taken for
hemoglobin assessment with a portable Hemocue photometer
(Hemocue AB, Ängelholm, Sweden). The remaining blood was
immediately transported to the central field laboratory for centrifu-
gation (1800  g, 10 min, room temperature) and serum separation
and then frozen at 20 C until transportation 2 times weekly to the
central serum storage location. Serum samples were kept at 70 C
until July 1999, when they were transported for further biochemical
analyses. Serum zinc and serum copper were assessed in duplicate
by atomic spectrometry (22). SF was analyzed by radioimmunoas-
say (Diagnostic Products, San Diego) and sTfR by enzyme-linked
immunosorbent assay (Ramco, Houston). To exclude the possible
influence of ongoing inflammation on the hematologic and bio-
chemical indexes, infants with fever or other symptoms of ongoing
infection had their blood sampling delayed for 2 wk.
Analytic approach and statistical methods
The analysis aimed at 1) assessment of any differentials in base-
line characteristics between treatment groups and between partic-
ipating and lost-to-follow-up subgroups; 2) intent-to-treat analy-
sis, ie, assessment of the effectiveness of the single or combined
iron and zinc treatment regimens on hematologic and biochemical
outcomes; 3) assessment of differences in adherence to the treat-
ment regimens between the groups, potential confounding, and
adjusted effectiveness analysis; 4) dose-effect analysis for the rel-
evant treatment groups and evaluation of whether a plateau in the
dose-effect curve could be observed; and 5) assessment of whether
the maximum obtained effect differed between the single and
combined supplements.
All statistical computations are based on the 549 infants with
both baseline and endpoint hematologic and biochemical results,
except for sTfR, for which the analysis is based on 507 infants.
885
Downloaded from https://academic.oup.com/ajcn/article-abstract/77/4/883/4689762 by guest on 12 March 2019
Non-normally-distributed variables, ie SF, sTfR, and serum zinc
values, were log-transformed (zero SF values were changed to
0.01). Main outcomes are shown as means and SDs or antilog of
mean logarithmic values, when applicable, and proportions.
Analysis of variance (ANOVA) was used in the intent-to-treat
analysis for the continuous variables comparing the study groups.
We also performed factorial analysis (two-factor ANOVA) with
supplementation with iron or supplementation with zinc to exam-
ine main effects and interactions on main outcomes. In the case of
a significant interaction between iron and zinc treatment, sub-
group analyses were performed and the one-sided P values were
Bonferroni corrected. For the iron outcomes (ie, hemoglobin, SF,
and sTfR), the Fe group was compared with the Fe+Zn and
placebo groups and the Fe+Zn group with the placebo group. For
the zinc outcomes (ie, serum Zn), the Zn group was compared
with the Fe+Zn and placebo groups and the Fe+Zn group with the
placebo group. The intent-to-treat analysis was repeated after
stratification for sex. Total supplementation volume was com-
pared between treatment groups, as well as side effects that could
influence adherence or outcome. Adjustments were thereafter
made for total volume, vomiting, and initial level of the outcome
parameter (for hemoglobin, SF, and sTfR but not for serum zinc
because of the absence of association between baseline and end-
point values).
Lowess smoothed plots were used to visualize the dose-effect
relation between volume taken and outcome for the different
treatment groups. This was followed by multivariate linear regres-
sion analyses to statistically assess the formulated hypotheses and
observations based on the Lowess curves. Adjustments were
made for potential confounding of the dose effect. Potential con-
founding was identified as cofactors with a P < 0.20 for any lin-
ear or nonlinear association with outcomes, treatment groups, or
total volume of treatment. In the regression analysis, presence of
vomiting reported as a side effect at the follow-up interview with
the mother, educational level of the mother, water source of the
household, and birth weight were identified as possible con-
founders and thus were included in the model. Confounding by
the measured variables was discarded as an important contribu-
tion to the main effects if the interactions or nonlinear introduc-
tions did not change the parameters describing the dose-effect by
> 10%. SPSS for WINDOWS 10 (SPSS Inc, Chicago) was used
for all statistical computations.
886 LIND ET AL

TABLE 1
Baseline characteristics of the participating infants1
Fe group Zn group Fe+Zn group Placebo group
(n = 136) (n = 134) (n = 136) (n = 143)
Household characteristics
No. of persons per household2 4 (1–10) 4 (2–9) 4 (2–10) 4 (1–9)
More than one child aged < 5 y (%) 26 25 31 29
Water source outside house (%) 24 31 31 27
Maternal characteristics
Age (y)3 29.8 ± 4.8 28.9 ± 4.7 29.5 ± 4.7 29.1 ± 5.1
Elementary education or more (%) 59 50 58 58
Infant characteristics
Girls [n (%)] 70 (52) 66 (49) 65 (48) 62 (43)
Birth weight (g)3 3198 ± 440 3210 ± 460 3220 ± 473 3196 ± 477

Downloaded from https://academic.oup.com/ajcn/article-abstract/77/4/883/4689762 by guest on 12 March 2019

Age at treatment start (mo)3 6.1 ± 0.4 6.1 ± 0.5 6.1 ± 0.4 6.1 ± 0.4
Weight at 6 mo (kg)3 7.17 ± 0.90 7.26 ± 0.91 7.30 ± 0.86 7.29 ± 0.93
Length at 6 mo (cm)3 65.5 ± 2.4 65.4 ± 2.3 65.5 ± 2.0 65.3 ± 2.8
Breast-fed at 6 mo [n (%)] 133 (98) 127 (95) 134 (99) 136 (96)
Baseline biochemistry indexes
Hemoglobin (g/L)3 114.0 ± 13.8 114.1 ± 13.6 112.1 ± 11.5 114.1 ± 14.5
Serum ferritin (g/L)4 37.2 ± 2.5 31.3 ± 2.7 33.9 ± 2.3 32.2 ± 2.7
sTfR (mg/L)4,5 6.83 ± 1.44 6.98 ± 1.44 7.28 ± 1.36 7.23 ± 1.33
Serum zinc (mol/L)4 8.94 ± 1.26 9.11 ± 1.32 8.99 ± 1.28 9.03 ± 1.25
Serum copper (mol/L)3 17.2 ± 4.8 17.3 ± 4.3 17.9 ± 4.4 17.2 ± 4.2
1
sTfR, serum transferrin receptor. There were no significant differences between study groups.
2
Median; range in parentheses.
3–
x ± SD.
4
Antilog of mean ln value ± SD.
5
n = 125, 123, 125, and 134 for the Fe, Zn, Fe+Zn, and placebo groups, respectively.

RESULTS
Six hundred eighty eligible infants were randomly assigned
to the 4 treatment groups (Figure 1); 666 (98%) completed the
6-mo course of supplementation, and 549 (81%) were assessed
with the use of both baseline and endpoint blood samples.
Baseline characteristics, including hemoglobin concentration
and iron and zinc status, were similar in the 4 treatment groups
(Table 1). Baseline biochemistry (hemoglobin, SF, sTfR, and
serum zinc) and sociodemographic characteristics did not dif-
fer significantly between those with and without complete bio-
chemical data (data not shown). At baseline, the infants gen-
erally had a satisfactory weight and length distribution; only
5% were stunted (length for age < 2 SDs of World Health
Organization growth reference values). Anemia (hemoglobin
< 110 g/L) was observed in 41%, low SF (< 12 g/L) in 15%,
iron deficiency anemia (low hemoglobin and SF) in 8%, and low
serum zinc (< 10.7 mol/L) in 78% of the infants.
Trial effectiveness (intent-to-treat analysis)
The Fe group had a significantly higher hemoglobin concentration
at the endpoint than did the placebo (difference between means: 5.9 g/L)
and Fe+Zn (difference: 4.1 g/L) groups (Table 2). The mean hemo-
globin concentration in the Fe+Zn group was not significantly different
from that in the placebo group. The prevalence of anemia (hemoglobin
< 110 g/L) at the endpoint was 34/136 (25%), 48/134 (36%), 51/136
(38%), and 63/143 (44%) in the Fe, Zn, Fe+Zn, and placebo groups,
respectively. The prevalence of anemia in the Fe group was significantly
different from that in the placebo (P < 0.001) and Fe+Zn (P = 0.026)
groups. However, the prevalence of anemia in the Fe+Zn group was not
significantly different from that in the placebo group (P = 0.266).

TABLE 2
Outcome of treatment (intent-to-treat analysis) on hemoglobin and measures of iron, zinc, and copper status1
Fe group Zn group Fe+Zn group Placebo group
(n = 136) (n = 134) (n = 136) (n = 143) P value2
Hemoglobin (g/L) 3
119.4 ± 15.3 4,5
115.7 ± 15.2 115.3 ± 13.9 113.5 ± 16.0 0.012
Serum ferritin (g/L)6 46.5 ± 2.04,5 13.3 ± 3.6 32.3 ± 2.94 12.9 ± 3.7 < 0.001
sTfR (mg/L)6,7 6.71 ± 1.334 9.78 ± 1.42 7.56 ± 1.364 9.02 ± 1.73 < 0.001
Serum zinc (mol/L)6 8.76 ± 1.24 11.58 ± 1.414 10.80 ± 1.344 9.06 ± 1.27 < 0.001
Serum copper (mol/L)3 15.2 ± 4.8 15.0 ± 5.1 14.7 ± 4.5 15.2 ± 5.1 0.814
1
sTfR, serum transferrin receptor.
2
ANOVA (on ln values for serum ferritin, sTfR, and serum zinc).
3–
x ± SD.
4
Significantly different from placebo, P < 0.05 (Bonferroni corrected).
5
Significantly different from Fe+Zn group, P < 0.05 (Bonferroni corrected).
6
Antilog of mean ln value ± SD.
7
n = 125, 123, 125, and 134 for the Fe, Zn, Fe+Zn, and placebo groups, respectively.
 IRON AND ZINC SUPPLEMENTATION IN INFANTS
TABLE 3
Details of follow-up at the end of supplementation
Fe group
(n = 136)
Treatment
Total supplement volume (mL)1,2 217 ± 513
Health, growth
Breast-fed at 12 mo [n (%)] 130 (96)
Weight at 12 mo (kg)1 8.33 ± 1.02
Length at 12 mo (cm)1 72.5 ± 3.0
Illness 2 wk before endpoint [n (%)] 38 (28)
Body temperature at endpoint (°C)1 36.8 ± 0.4
Side effects
Any perceived side effect [n (%)] 81 (60)
Vomiting [n (%)]4 42 (31)
1–
x ± SD.
2
ANOVA P < 0.05 for difference between study groups.
3
Significantly different from Zn and placebo groups, P < 0.05 (Bonferroni corrected).
4
Defined as vomiting reported as a side effect during supplementation at the follow-up interview.
5
Significantly different from all other groups, P < 0.05.
Similarly, the Fe and Fe+Zn groups had significantly higher SF
and lower sTfR concentrations than did the placebo group
(Table 2). The effect in the Fe group was significantly higher than
in the Fe+Zn group regarding SF (difference between geometric
means: 14.2 g/L). At the endpoint, the prevalence of low SF (SF
< 12 g/L) was 7/136 (5%), 47/134 (35%), 14/136 (10%), and 54/
143 (37%) in the Fe, Zn, Fe+Zn, and placebo groups, respectively.
The prevalence of iron deficiency anemia (hemoglobin < 110 g/L
and SF < 12 g/L) was 3/136 (2%), 21/134 (16%), 5/136 (3%),
and 30/143 (21%) in the Fe, Zn, Fe+Zn, and placebo groups,
respectively. The Fe and Fe+Zn groups had significantly lower
prevalences of iron deficiency anemia than did the placebo group
(both P < 0.001). Significant differences in iron deficiency
anemia prevalence could not be shown between the Zn and
placebo groups (P = 0.255), nor between the Fe and Fe+Zn
groups (P = 0.473). The prevalence of iron deficiency anemia
increased significantly from 6 to 12 mo of age in the placebo
group (P = 0.002). Two-factor ANOVA showed a significant
interaction between iron and zinc treatment for both hemoglobin
(P = 0.021) and SF (P = 0.032) but not for sTfR.
The Zn and Fe+Zn treatment groups had higher mean serum
zinc concentrations at the endpoint than did the placebo group.
The proportion of low serum zinc (< 10.7 mol/L) concentrations
at the endpoint was 118/136 (87%), 63/134 (47%), 74/136 (54%),
and 111/143 (78%) in the Fe, Zn, Fe+Zn, and placebo groups,
respectively. The Zn and Fe+Zn groups had significantly fewer
infants with low serum zinc concentrations than did the placebo
(both P < 0.001) or Fe (both P < 0.001) group. Two-factor
ANOVA showed no significant interaction between iron and zinc
treatment on serum zinc levels.
There were no significant differences in treatment effects
between boys and girls regarding hemoglobin, SF, or serum zinc
outcomes (data not shown). The 4 treatment groups did not differ
significantly in mean serum copper concentrations at the endpoint
(Table 2).
Deviation from protocol and side effects
The total intake of supplements was lower in the Fe+Zn group
than in the Zn (mean difference: 29 mL; P < 0.001) and placebo
groups but was not significantly different from that in the Fe group
887
Zn group Fe+Zn group Placebo group
(n = 134) (n = 136) (n = 143)
236 ± 52 207 ± 673 242 ± 50
123 (92) 127 (93) 134 (94)
8.48 ± 0.97 8.40 ± 0.94 8.31 ± 0.94
72.4 ± 2.8 72.3 ± 2.4 72.4 ± 2.8
39 (29) 36 (27) 38 (27)
36.8 ± 0.4 36.7 ± 0.4 36.8 ± 0.4
84 (63) 97 (82) 82 (57)
Downloaded from https://academic.oup.com/ajcn/article-abstract/77/4/883/4689762 by guest on 12 March 2019
46 (34) 72 (53)5 39 (27)
(P = 0.129; Table 3). Vomiting was more common in the Fe+Zn
group (Table 3). Other perceived side effects (abdominal pain,
diarrhea, constipation, poor appetite, increased crying, and fussi-
ness) did not differ significantly between treatment groups. Vom-
iting was negatively associated with the total volume of supple-
ment given (r = –0.387, P < 0.001), which also implies the
possibility of loss of supplement after the daily dose.
Baseline and endpoint values were associated for hemo-
globin, SF, and sTfR but not for serum zinc. Consumption of
supplement and reported vomiting were, as noted, signifi-
cantly different between treatment groups. Consequently,
adjusting for these factors slightly changed the estimated
effect size for serum zinc at the endpoint, and the contrast
result (Zn compared with Fe+Zn) was no longer significant
(geometric mean of serum zinc: 11.51 and 10.99 mol/L,
respectively, P = 0.185).
Efficacy of the Fe, Zn, and Fe+Zn regimens
Serum ferritin outcome
To evaluate the efficacy of the Fe and Fe+Zn regimens on
iron status (SF) at the endpoint, the dose effect over the range
of total supplement intake during the treatment period was stud-
ied (Figure 2). Lowess curves for the Fe and Fe+Zn groups indi-
cated an initial dose effect, followed by a plateau with no further
increased effect. In the ordinary least-squares regression model,
the dose effect up to 200 mL was compared with that above 200 mL
of total supplementation (Table 4). Adjustments for potential con-
founding covariates were made if fulfilling the set criteria. In the
Fe group, there was a significant dose effect up to 200 mL (cor-
responding to a total dose of 1250 mg Fe over a period of 6 mo,
or 7 mg/d). After that total dose, no further increase in effect was
shown. Above 200 mL of total supplement, the effect in the Fe
group was significantly higher than that in the Fe+Zn group
( = 0.171, P = 0.003), but there was no significant interaction
between treatment group and volume. Despite the appearance of
the Lowess curves, no significant difference in effect for a given
dose and no interaction in dose effect between the groups for sup-
plementation amounts < 200 mL could be shown between the Fe
and Fe+Zn groups (data not shown).
888 LIND ET AL

Downloaded from https://academic.oup.com/ajcn/article-abstract/77/4/883/4689762 by guest on 12 March 2019

FIGURE 2. Volume of supplement consumed from 6 to 12 mo and
serum ferritin concentrations (logarithmic scale) at the endpoint. Shown
are Lowess curves for the 4 treatment groups (n = 549).
FIGURE 3. Volume of supplement consumed from 6 to 12 mo and
serum zinc concentrations at the endpoint. Shown are Lowess curves for
the 4 treatment groups (n = 549).

status of daily iron supplementation and that a daily dose of 10 mg

Serum zinc outcome
In the Zn group, a significant dose-effect relation was shown
(Figure 3), without any evident plateau and without any identi-
fied confounding ( for consumed supplement in mL: 0.0024,
P = 0.001, r2 = 0.15). In the Fe+Zn treatment group, there was a
significant dose effect for supplementation volumes < 200 mL
( = 0.0020, P = 0.023) but not for volumes ≥ 200 mL. A signifi-
cant interaction between volumes < and ≥ 200 mL was not shown,
and the linear regression equation, including an interaction term,
was not significant. For supplementation amounts ≥ 200 mL, there
was a small but significant difference in effect levels between the
Zn and Fe+Zn groups ( = 0.097, P = 0.048).
DISCUSSION
This study of the effect of single or combined iron and zinc sup-
plementation in an iron- and zinc-deficient infant population pro-
vides evidence that a combined supplement is less efficacious than
are single supplements in improving iron and zinc status. Fur-
thermore, data are presented suggesting a negative effect on zinc
Fe may be unnecessarily high.
No deviations from the protocol were observed in the alloca-
tion of treatment groups and masking. The poorer adherence to
the prescribed total dose in the combined supplement group could
potentially bias the results, and the higher frequency of vomiting
in the same group could spuriously reduce the effect estimate.
However, adjustments for these factors modified the results only
slightly. Socioeconomic factors related to participation (supple-
ment amount) and biochemical outcomes might bias the dose-
effect analysis, and adjustments were made when warranted. No
major influence on effect estimates was found.
This group of infants generally had a fair nutritional status—
as expressed by anthropometry—at baseline, but had a relatively
high frequency of low iron and zinc status and anemia. Absolute
levels of effects would be different in other populations, but the
observed relative differences in effects between treatment regi-
mens as well as most of the dose-effect observations are most
likely relevant for other iron- and zinc-deficient infant populations
in the world.
Iron, when given alone, had a significant positive effect on iron
status as assessed by higher hemoglobin and SF concentrations,

TABLE 4
Dose effect on iron status (efficacy analysis)
Outcome: ln serum ferritin at endpoint Fe P value Fe+Zn1 P value
Constant 1.214 0.149 0.409 0.586
Consumed supplement (mL) 0.00975 0.036 0.0144 0.001
ln Serum ferritin at baseline 0.210 0.001 0.287 0.006
Volume group (1 = ≥ 200 mL, 0 = < 200 mL) 1.958 0.059 3.364 0.006
Consumed supplement  volume group 0.00951 0.078 0.0189 0.001
F 6.29 < 0.001 7.05 < 0.001
r2 0.174 0.267
1
Adjusted for educational level and educational level  volume group.
 IRON AND ZINC SUPPLEMENTATION IN INFANTS
lower sTfR, and a lower prevalence of anemia than in all the other
groups. It is apparent that treatment with the combination of iron
and zinc was less effective in improving iron status: the high
prevalence of anemia at 6 mo of age had not changed by 12 mo,
and both hemoglobin and SF were significantly lower than in the
Fe group. The combined Fe+Zn supplement resulted in some
improvement in iron status, however: SF was higher and sTfR
lower than in the placebo group. Thus, the addition of zinc to the
iron supplement impaired the utilization of iron but not to the
extent that it had no effect on iron status. Finally, in the Zn
group the sTfR estimate was higher, although not significantly
so (P = 0.097); if correct, this may indicate increased intracellu-
lar iron needs in that group.
It is likely that zinc interfered with the absorption of iron, as
was shown previously for these elements when given in water
solution to adults (14, 15). The mechanisms regulating iron
absorption have been discovered (23). An iron transporter at the
apical side of the enterocyte, the DMT1 (divalent metal transporter
1, also known as Nramp2 or DCT1), has been shown to also trans-
port zinc ions (24). Although this has not yet been shown in mam-
malian systems, it is possible that high concentrations of zinc may
interfere with the transport of iron into the intestinal cell, without
completely blocking it. However, the acute effects of zinc on iron
transport by DMT1, as well as possible long-term adaptations,
need further study.
Serum zinc is not an ideal indicator of zinc status because it is
affected by several factors other than zinc intake and zinc status
(25). It is generally agreed, however, that serum zinc is of some
value as an indicator when used for larger groups in similar settings
and when care is taken to standardize conditions for blood sam-
pling. The Zn group had significantly higher serum zinc concen-
trations than did the placebo group and higher concentrations than
did the Fe+Zn group, although this difference was not significant
(P = 0.057). This may suggest that zinc alone was better utilized
than when given in combination with iron. However, zinc in the
combined Fe+Zn supplement obviously was utilized to some extent,
because serum zinc was higher in that group than in the placebo
group, which received no supplemental zinc. Furthermore, the Fe
group had a significantly higher proportion of low serum zinc val-
ues than did the placebo group. Our findings are thus in agreement
with studies in adults showing that high concentrations of inorganic
iron inhibit zinc absorption when given in water solution (11–13).
The cellular mechanisms for zinc uptake and transport have
been discovered, and some of these transporters, such as ZIP-1,
ZnT-1, and ZnT-4, have been found in the small intestine (26,
27) The precise roles of these zinc transporters in the cellular
uptake of zinc at the apical side, in intracellular transport, or in
efflux at the basolateral side have not been delineated, and it is
not known whether they are affected by high amounts of iron. It
is obvious that further studies on the interaction between iron
and zinc are needed.
High dietary intake of iron has been shown to affect copper
status in some studies (17, 28). However, these studies found
effects when iron was added to infant formula and not when
given as a supplement. We found no negative effects of iron
supplements alone or in combination with zinc on serum cop-
per. It is possible that the ratio of iron to dietary copper was
not high enough to have an effect or, perhaps more likely, that
iron given apart from a meal did not affect copper absorption
from the diet. High dietary intakes of zinc have also been
shown to affect copper absorption and status in experimental
889
animals (29), and zinc supplementation in addition to food-
based nutrition rehabilitation may reduce plasma copper con-
centrations in very malnourished children with persistent diar-
rhea (30). It is likely that these 2 elements interact only when
given together in a meal and not, as in our study, when zinc is
given apart from meals.
One-half of the parents reported vomiting due to the Fe+Zn
supplement. The daily morbidity registration showed that 37% in
the Fe+Zn group but only 20% in the placebo group experienced
vomiting without concurrent illness. The reason for this increased
rate of vomiting without other differences in abdominal symptoms
is unclear.
Our results suggest that single supplements of iron and zinc
Downloaded from https://academic.oup.com/ajcn/article-abstract/77/4/883/4689762 by guest on 12 March 2019
were more effective in improving iron and zinc status, respec-
tively, than was a combination of the 2, even after the analysis was
controlled for compliance and possible side effects. We also
believe that our data show that this population of infants needs
increased intake of both micronutrients. Whether ratios of the 2
elements in a supplement other than the 1:1 we used are more
effective in improving both iron and zinc status is not known, and
this was not addressed in our study. Note that an increase in the
dose of iron and zinc in a combined daily supplement would most
likely fail to compensate for the interaction because the dose-
effect curve leveled off below the single supplement.
On the basis of the results of animal experiments it has been
suggested that intermittent supplementation of iron may be
equally as or more effective than daily supplementation in pre-
venting iron deficiency, the so-called mucosal block theory (31).
However, experiments in humans have not provided evidence for
an increased efficacy with weekly supplementation (32). Thus, the
issue is rather whether an intermittent (eg, weekly) iron dose fre-
quency will supply the sufficient total dose. Our results indicate
that the total dose provided according to protocol (10 mg Fe/d)
was unnecessarily high. No further effect was obtained beyond a
total supplementation volume corresponding to 1250 mg Fe (pro-
tocol: 1800 mg) over the 6 mo of the study, or 7 mg/d. Whether
weekly iron supplements would reduce the inhibitory effect of
iron on zinc absorption is not known. It is possible that a bolus
dose may inhibit zinc absorption that same day, but it is unlikely
to have a long-lasting effect on zinc uptake.
It is highly likely that zinc supplements need to be given every
day or most days, because there are no stores of zinc and zinc
turnover is rapid. No plateau was observed in the dose-effect
analysis of the Zn group. However, the analysis was done with
serum zinc as the outcome. Whether the dose-effect pattern is sim-
ilar in relation to functional outcomes needs to be assessed.
To summarize, our study has provided evidence that combined
iron and zinc supplementation is not optimal. The combined sup-
plement showed no effect on hemoglobin, whereas the single iron
supplement had a clinically and statistically significant effect. The
combined supplement also had a significantly lower effect than
did the single supplements on SF. Furthermore, evidence is pro-
vided that single daily iron supplementation inhibits zinc absorp-
tion from food. In situations in which a supplementation strategy
is more feasible than a food-based approach, innovative dose reg-
imens for the provision of these micronutrients are needed.
We are grateful to the families in Purworejo who participated in the trial; to
Sunarti, Weng-In Leong, and Michael Crane, who did an excellent job with the
biochemical analyses; and to the dedicated field staff of the CHN-RL at Pur-
worejo and Yogyakarta, without whom this effort would have been impossible.
890 LIND ET AL
TL was the main author of the paper and also participated in the planning,
performance, and data analysis of the trial. BL participated in the study design,
data analysis, and writing of the manuscript. HS and E-CE contributed to the
data analysis and writing of the manuscript. DI took part in the study design
and data collection. RS took part in the data collection. L-ÅP participated in
designing the study, analyzing the data, and writing the manuscript. None of
the contributing authors had any financial or personal interests in any of the
bodies sponsoring this research.
REFERENCES
1. DeMaeyer E, Adiels-Tegman M. The prevalence of anaemia in the
world. World Health Stat Q 1985;38:302–16.
2. ACC/SCN. Third report on the world nutrition situation. Geneva:
Administrative Committee on Coordination/Sub-committee on Nutri-
tion, 1997.
3. Murray C, Lopez A. The global burden of disease. Boston: Harvard
University Press, 1996.
4. Black R. Micronutrient deficiencies. Global Forum for Health
Research, 1999. Internet: http://www.globalforumhealth.org/pages/
index.asp (accessed 21 Jan 2003).
5. Lonnerdal B. Dietary factors influencing zinc absorption. J Nutr
2000;130(suppl):1378S–83S.
6. Brown KH, Peerson JM, Allen LH. Effect of zinc supplementation
on children’s growth: a meta-analysis of intervention trials. Bibl Nutr
Dieta 1998;102:76–83.
7. Zinc Investigators’ Collaborative Group. Prevention of diarrhea and
pneumonia by zinc supplementation in children in developing coun-
tries: pooled analysis of randomized controlled trials. J Pediatr 1999;
135:689–97.
8. Shankar AH, Genton B, Baisor M, et al. The influence of zinc sup-
plementation on morbidity due to Plasmodium falciparum: a ran-
domized trial in preschool children in Papua New Guinea. Am J Trop
Med Hyg 2000;62:663–9.
9. Müller O, Becher H, van Zweeden AB, et al. Effect of zinc supple-
mentation on malaria and other causes of morbidity in west African
children: randomised double blind placebo controlled trial. BMJ
2001;322:1567.
10. The Zinc Against Plasmodium Study Group. Effect of zinc on the
treatment of Plasmodium falciparum malaria in children: a random-
ized controlled trial. Am J Clin Nutr 2002;76:805–12.
11. Solomons NW, Jacob RA. Studies on the bioavailability of zinc in
humans: effects of heme and nonheme iron on the absorption of zinc.
Am J Clin Nutr 1981;34:475–82.
12. Valberg LS, Flanagan PR, Chamberlain MJ. Effects of iron, tin, and
copper on zinc absorption in humans. Am J Clin Nutr 1984;40:
536–41.
13. Sandstrom B, Davidsson L, Cederblad A, Lonnerdal B. Oral iron,
dietary ligands and zinc absorption. J Nutr 1985;115:411–4.
14. Crofton RW, Gvozdanovic D, Gvozdanovic S, et al. Inorganic zinc
and the intestinal absorption of ferrous iron. Am J Clin Nutr 1989;
50:141–4.
15. Rossander-Hulten L, Brune M, Sandstrom B, Lonnerdal B, Hallberg L.
Competitive inhibition of iron absorption by manganese and zinc in
humans. Am J Clin Nutr 1991;54:152–6.
16. Davidsson L, Mackenzie J, Kastenmayer P, et al. Dietary fiber in
weaning cereals: a study of the effect on stool characteristics and
absorption of energy, nitrogen, and minerals in healthy infants. J Pedi-
atr Gastroenterol Nutr 1996;22:167–79.
17. Haschke F, Ziegler EE, Edwards BB, Fomon SJ. Effect of iron forti-
fication of infant formula on trace mineral absorption. J Pediatr Gas-
troenterol Nutr 1986;5:768–73.
18. Fairweather-Tait SJ, Wharf SG, Fox TE. Zinc absorption in infants
fed iron-fortified weaning food. Am J Clin Nutr 1995;62:785–9.
19. Kolsteren PW. Growth faltering in Madura, Indonesia: a comparison
with the NCHS reference and data from Kasongo, Zaire. Ann Trop
Paediatr 1996;16:233–42.
20. Kodyat B, Kosen S, de Pee S. Iron deficiency in Indonesia: current
Downloaded from https://academic.oup.com/ajcn/article-abstract/77/4/883/4689762 by guest on 12 March 2019
situation and intervention. Nutr Res 1998;18:1953–63.
21. Ninuk TSH, Dibley MJ, Sadjimin T, Serdula M. Food and nutrient
intakes of infants and young children in Central Java, Indonesia.
Yogyakarta, Indonesia: Clinical Epidemiology and Biostatistics Unit,
University of Gadjah Mada, 1997.
22. Clegg M, Keen C, Lönnerdal B, Hurley L. Influence of ashing tech-
niques on the analysis of trace elements in animal tissue. I. Wet ash-
ing. Biol Trace Elem Res 1981;3:107–15.
23. Eisenstein RS, Blemings KP. Iron regulatory proteins, iron
responsive elements and iron homeostasis. J Nutr 1998;128:
2295–8.
24. Gunshin H, Mackenzie B, Berger UV, et al. Cloning and characteri-
zation of a mammalian proton-coupled metal-ion transporter. Nature
1997;388:482–8.
25. Brown KH. Effect of infections on plasma zinc concentration and
implications for zinc status assessment in low-income countries. Am
J Clin Nutr 1998;68(suppl):425S–9S.
26. McMahon RJ, Cousins RJ. Regulation of the zinc transporter ZnT-1
by dietary zinc. Proc Natl Acad Sci U S A 1998;95:4841–6.
27. Cousins RJ, McMahon RJ. Integrative aspects of zinc transporters. J
Nutr 2000;130(suppl):1384S–7S.
28. Lonnerdal B, Hernell O. Iron, zinc, copper and selenium status of
breast-fed infants and infants fed trace element fortified milk-based
infant formula. Acta Paediatr 1994;83:367–73.
29. Wapnir RA, Lee SY. Dietary regulation of copper absorption and stor-
age in rats: effects of sodium, zinc and histidine-zinc. J Am Coll Nutr
1993;12:714–9.
30. Bhutta ZA, Nizami SQ, Isani Z. Zinc supplementation in malnour-
ished children with persistent diarrhea in Pakistan. Pediatrics [serial
online] 1999;103:e42. Internet: http://www.pediatrics.org/cgi/content/
full/103/4/e42 (accessed 21 Jan 2003).
31. Viteri FE, Liu X, Tolomei K, Martin A. True absorption and reten-
tion of supplemental iron is more efficient when iron is administered
every three days rather than daily to iron-normal and iron-deficient
rats. J Nutr 1995;125:82–91.
32. Beaton G, McCabe G. Efficacy of intermittent iron supplementation
in the control of iron deficiency anaemia in developing countries: an
analysis of experience. Ottawa: The Micronutrient Initiative, 1999.